A retrospective study based on SEER database: not all high-risk factors are equal for stage II colon cancer

Abstract

Background: For stage II colon cancer, understanding of high-risk factors (HRFs) that affect the overall survival (OS) and the benefit of chemotherapy is limited. Meanwhile, no stable predictor can effectively predict OS of stage II colon cancer to date. Our study is aimed to identify HRFs associated with OS of stage II colon cancer, to quantify the risk conferred by each HRF, and to evaluate OS benefit gained by chemotherapy. Meanwhile, we attempt to establish a nomogram model for stage II colon cancer. Methods:The clinical variables of patients with stage II colon cancer between 2000 and 2018 were retrieved from the Surveillance, Epidemiology, and End Results (SEER) database. Univariate and multivariate analysis were performed to filtered out all the HRFs. We calculated the hazard ratios (HR) and evaluated the survival benefit of adjuvant chemotherapy for each HRF and combinations of HRFs. Then, a nomogram model based on all HRFs was established and verified.Results: A total of 39,103 patients with stage II colon cancer were included. T4b tumors were the highest risk for reduced OS [HR =2.821; 95% confidence interval (CI): 1.949-4.082], mucin-producing tumors (HR =2.412; 95% CI: 1.326-4.388) the second, and lymph node (LN) examined less than 12 (HR =2.200; 95% CI: 1.786-2.710) the third. T4 tumors (HR =0.790; 95% CI: 0.542-1.151), poorly/undifferentiated tumors (HR =0.468; 95% CI: 0.237-0.924), and some combinations of HRFs containing either could benefit from adjuvant chemotherapy. Meanwhile, we established an effective nomogram model based on the identified HRFs. Conclusions:The study has identified several novel HRFs for stage II colon cancer. Adjuvant chemotherapy has considerable OS benefit for stage II colon cancers with some specific HRFs, and treatment plans need to be individualized. Type and number of HRFs should be taken into consideration when recommending adjuvant chemotherapy. Our new nomogram model has better predictive ability and stability than the tumor-node-metastasis (TNM) stage system of American Joint Committee on Cancer (AJCC) staging system.

Can clinicopathologic high-risk features in T3N0 colon cancer be reliable prognostic factors?

“…A study with the SEER (Surveillance, Epidemiology, and End Results) database showed that OS decreased according to the number of HRFs; 78.4% for patients with no HRF, 71.99% for those with one HRF, and 59.24% for those with more than 2 HRFs, which is worse than the OS of stage III disease, which is reported to be 64.68% [ 23 ]. The detrimental effect of multiple HRFs on OS was also demonstrated in another study using data from the SEER database from a different period [ 14 ]. The International Duration Evaluation of Adjuvant Chemotherapy Collaboration data also reported that the 5-year disease-free survival was 74.8% for stage II patients with 2 or more HRFs, compared with 87.3% for patients with one HRF [ 25 ].…”

“…The prognostic benefit of adjuvant chemotherapy in stage II cancer has been controversial in many previous reports. Some reported that a clear OS benefit of adjuvant chemotherapy had not been shown for stage II colon cancer [ 1 14 15 16 17 ], and some studies reported higher survival outcomes of stage II colorectal cancer with adjuvant chemotherapy in subgroups of patients [ 17 18 19 ]. Studies reporting prognostic improvement with adjuvant chemotherapy indicate that this benefit was mainly seen in the subgroup of patients with T4 disease [ 4 20 21 ].…”

“…Low compliance with the adjuvant chemotherapy administration guidelines in stage II disease may be a contributing factor. In many reports, adjuvant chemotherapy was given to less than 40% of patients with stage II disease, and even patients without HRF received adjuvant chemotherapy in around 10% of cases [ 14 17 23 26 27 ]. In the current study, adjuvant chemotherapy was given at a greater proportion than in previous reports.…”

“…Another factor is that HRF included in various studies differed. Some studies did not include the recommended HRF [ 14 23 ], while some included other clinicopathologic features such as location, age, microsatellite instability status, and tumor deposit [ 17 29 ].…”

“…The small proportion of patients with these HRFs may contribute toward inconsistent reporting. The proportion of patients with <12 LNs was reported to be approximately 10% [ 14 17 ]. The proportion with margin involvement and poor differentiation was also very low [ 14 17 23 ].…”